AI engineering author, Stanford instructor
Chip Huyen
Profile
Chip Huyen is the author of two of the most-read books in practical ML and AI today — Designing Machine Learning Systems (2022) and AI Engineering (2025). If you’re trying to go from “I’ve made an LLM return something” to “I run this in production and it doesn’t fall over,” her books are where most people land.
She grew up in Vietnam, studied at Stanford, and stayed on to teach CS 329S: Machine Learning Systems Design — a project-heavy course whose lecture notes became the first book. Before that she built ML tooling at NVIDIA, Netflix, and Snorkel AI. In 2022 she co-founded Claypot AI to do real-time ML infrastructure; in early 2024 Voltron Data acquired the company, and she joined as VP of AI and Open Source, working on GPU-native data processing on top of Apache Arrow and Ibis.
What makes her worth reading is that she refuses to write hype. Her blog posts — the 200-ML-tools landscape, the real-time ML series, the 2025 “common pitfalls” piece — are surveys of what people are actually doing, with opinions attached. She treats evaluation, data, and infrastructure as the hard parts, because they are. For someone learning AI, the shortest path to understanding how real systems get built is to read her cover to cover and then go build something.
She also writes outside tech: four bestselling Vietnamese travel books, starting with Xách Ba Lô Lên Và Đi. The writing voice carries over — plain, observational, no posturing.
Books
AI Engineering: Building Applications with Foundation Models The practical handbook for building products on top of foundation models — evaluation, RAG, fine-tuning, agents, inference optimization, and the rest of the stack. Designing Machine Learning Systems The predecessor and companion to AI Engineering — how to design, deploy, and maintain ML systems that survive contact with production data.Key Articles & Papers
AI Engineering (book companion repo) Common pitfalls when building generative AI applications Building LLM applications for production Real-time machine learning: challenges and solutions Machine learning is going real-time What I learned from looking at 200 machine learning tools MLOps guide Open Source LLM Tools (llama-police)Spotify Podcasts
